Low cost High Performance Wireless Mesh Networks (2 students)


Much research needs to be done towards realizing high throughput in

wireless mesh network environments. Current mesh networks deliver poor

throughput in multi-hop settings. We are currently investigating

different research challenges in the space of multi-radio and

multi-channel mesh networks. In 715 Broadway, we have deployed a 12 node

wireless mesh network where each node has multiple radios. This testbed

is supposed to be a multi-radio multi channel testbed where one can

potentially set different rafios to different channels to improve

throughput. Currently, we have 2-3 students working on this project. As

part of the course project, the students would be performing a detailed

measurement student in this testbed to instrument the performance under

different configurations. The goal is to use this measurement to

determine the design principles for constructing high performance

multi-radio networks.


Cell-phone based Health care records system (2 students)


This project would involve developing a simple electronic medical system

(note, a simple database not a fullfledged EMR system) in a programmable

cellphone environment. Here, the goal is to have a setting where

healthworkers in the field collect health records on a cell phone and

these records are synced up with each other using SMS. The only means of

communication between the server and the cellphone will be an SMS

message. The challenges are how one would compress health care records

into small set of SMS messages, how one would maintain local state to

propagate Delta changes (instead of entire records), how to search

efficiently across cellphones etc.


Cell-phone based Financial Transaction System (2 students)


A cell phone based financial transaction system is supposed to provide

the flexibility of executing financial transactions in an intermittent

environment with lazy updates using SMS with the main server.

One student, Michael Paik, is working on a cellphone based financial

transaction system to support ATMs and Microfinance applications which

supports "low-cost authentication" and the ability

to provide "offline transactions" for financial transactions. Here

again, the means of synchronization between the server and the clients

would involve using only SMS. The student will work with Mike in the

development of the system.


Automated Diabetetic Retinopathy Detection (2 students)


One student, Koray, is working on an automated system to detect diabetic

retinopathy from retinal images using machine learning algorithms. The

basic system has been built - however several engineering

mechanisms + configuration testing is essential to make the system

robust and perfect. The current system still has a high false positive

ratio. The student will work in Koray in improving the robustness of the

algorithm and reduce the false positive ratio of the detection system.


Low cost paper Watermarking (1-2 students)


In the previous semester, a few students build a basic approach towards

extracting speckles in paper to show how one can watermark paper using a

simple USB microscope. The student(s) will work with some members of the

group to improve the system to develop a full fledged watermarking system.


Medical Education Modules (3-4 students)


This projects which contains two parts involves the creation of medical

education modules for teaching health-care workers in rural places about

specific diseases. The specific focus for these medical modules would be

two topics: (a) malaria and infectious diseases; (b) obstetrics. Here,

the student(s) would take up a specific topic and work with experts in

NYU Medical School on the creation of medical education modules for

their chosen topic. This project will build upon the WiSe-MD that has

already been developed in NYU for surgical education.


Intermittent Content Distribution (2-3 students)


This project would involve building a simple content distribution

network that operates in intermittent environment. An intermittent link

is one that is off and on. The goal here is to build an intermittent

proxy that can intelligently implement "store and forward". Jay Chen, a

PhD student, is currentlyw orking on a generic intermittent proxy

implementation. The students working on this project will work with Jay

on extending his proxy implementation to support content distribution

network. The medical education modules developed by the other group will

be the content that will be used for testing for this intermittent CDN.


Low-bandwidth Video Streaming (2-3 students)


This project involves building a low-bandwidth video streaming system

that can be used for video conferencing and one-way video streaming

(like MPEG). The goal is to get the streaming to work over 64 Kbps and

22 Kbps GPRS links. Some of the challenges in building such a system

would involve: (a) need for a proactive recovery system that can recover

important packets in the stream over a lossy link;

(b) developing a video codec that is applicable for low bandwidth

environments; (c) doing optimizations in static video environments

(background roughly fixed). There is a team of 1-2 students already

working on this. The group would work with this team.


Traffic Signaling System (2 students)


Road traffic systems in developing countries are perenially congested.

David Bindel and I have been working on developing an alternate

signaling protocol that internetworks the traffic signals to improve the

overall throughput of the road system. This project would involve

building a traffic simulator that can test some of these ideas and

improving over them. If successful, we can get traffic data from

developing countries to test this hypothesis in greater detail.


User Interface Design (2 students)


Researcher in MSR India have developed a text-free interface for

semi-literate and illiterate people in developing countries. This

project would involve coming up with alternative user interface designs

(especially for cell phones) that can are usable for semi-literate or

illiterate people. This is a very good challenge for design students.

Developing the correct interface can be a very good research challenge.